盲卷积信号的一种高效分离算法

An Efficient Algorithm of Blind Source Separation for Convolved Signals

针对盲卷积信号提出一种高效分离算法,采用分段的思想,对每短时段信号通过检验分离效果而灵活选择是否需要重新训练分离矩阵参数,既考虑了混合模型的时变性,又大大提高了盲分离的效率。研究还提出并应用了限值检验控制准则,来有效进行迭代算法的控制和分离效果检验的判决,保证了算法的执行效率。仿真实验分析通过比较不分段盲分离算法和分段逐段盲分离算法,表明了该算法的高效性。

An efficient algorithm of blind source separation for convolved signals with the blocking idea was proposed, which could decide whether a new separation matrix was trained or not after detecting the separation results for every short signal. It not only thinks about the time-varying of the mixture model but also greatly improves the efficiency of BSS. The detection and control criteria based on the threshold was also proposed and effectively applied to the control of the iterative algorithm and the judgment of the separation effect. The simulation experiment shows that this algorithm is more efficient than the other algorithms.